vbo

Few conflicting thoughts:

Having third party embedded researchers red teaming alignment is OK.

The alternative is having a bureaucratic agency like the US FDA testing and vetting models. The government probably couldn't keep pace with AI research right now, but it's where we'll eventually end up at anyway.

Fable/Astra can generate massive income for years even with no further advances.

Dario, Sam and Elon would not agree to do this if they didn't think alignment is possible in the short term, and the pain incurred by third party evaluators would be limited. Surely, if recursive self improvement works wonders in AI research, it can similarly work wonders in alignment. So this could be a PR stunt.

P-doom not negligible indeed.

I think there may be more to overcomplicated brews than lifestyle or status. It's a desire to get technical in a world where so much of life has been automated away from the consumer's view or ability to affect it. Coffee is a way to tinker with something in the same way previous generations tinkered with their cars or radios or whatever. It's an outlet for creativity and technical skill building for those that engage in brewing.

Coffee shops will sell rituals, status, prestige, sophistication or the appearance thereof. Same as every other business. But that's not to say the product can't be superior - it can. It also doesn't say that every product that makes use of those marketing techniques is superior; but even if it isn't, if the customer walks away happy, they must've done something right, right?

Don't we do the same with the technology we're working on? How often is it truly better beyond any critique? People were getting stuff done even before our products were around with less fuss and a different set of problems. We do what we do to end up busy with stuff so we can do what we do all over again, don't we? I digress.

It does get tiresome when everyone is trying to sell you an experience and it becomes disappointing when the selling of experiences becomes so commoditized, the thing being sold loses its credibility as something special on account of being sold as such. Is it a crisis of authenticity?

To each, their own. I used to tinker with espresso based drinks, but I'm mostly over it. I've learned to discern (some) better coffee beans from others, but I mostly don't drink that - I can't justify paying that much for a coffee I brew myself and that I may botch out of being in a hurry. It's also a distraction that takes time I don't have anymore. But it was fun to explore for a while and I now own a very fancy looking espresso machine, grinder and all sorts of acccessories.

I'm working on a tool to generate and host full stack web apps from prompts (just like everyone else). I'm loving it. Using llms to do as much of my coding as possible, so in a way eating my own dog food, although it's a more developer-driven effort than what the end product will be.

Strange thing is, the most time consuming part of getting this ready for a user facing launch is not the code generating, but all the scaffolding/queues/storage to run it.

Looks good, might give it a try. I was looking for something similar to provide a unified interface for gpt and claude and eventually hacked something together myself, as none of the solutions I found could deal with structured output properly across both vendors.

I miss the 00s internet. I miss IRC and geeking out for the sake of it. Maybe i'm just missing my younger years, but I think there was a distinct feeling back then, of wonder and being amongst the first to tinker with these promising technologies that were going to change the world for the better and now it's 2024 and we've screwed it all up.

Point is we couldn't have a way of consenting to ai training until after we had llms. And I'm guessing we will, pretty quickly.

I don't want to defend Altman. He may or may not be a good actor. But as an engineer, I love the idea of building something magical, yet lately that's not straightforward tinkering - unless you force your way - because people raise all sorts of concerns that they wouldn't have 30 years ago. Google (search) was built on similar data harvesting and we all loved it in the early days, because it was immensely useful. So is ChatGPT, but people are far more vocal nowadays about how what it's doing is wrong from various angles. And all their concerns are valid. But if openai had started out by seeking permission to train on any and every piece of content out there (like this comment, for example) they wouldn't have been able to create something as good (and bad) as ChatGPT. In the early search days, this was settled (for a while) via robots.txt, which for all intents and purposes openai should be adhering to anyway.

But it's more nuanced for LLMs, because LLMs create derivative content, and we're going to have to decide how we think about and regulate what is essentially a new domain and method and angle on existing legislation. Until that happens, there will be friction, and given we live in these particular times, people will be outraged.

That said, using SJ's voice given she explicitly refused is unacceptable. It gets interesting if there really is a voice actor that sounds just like her, but now that openai ceased using that voice, the chances of seeing that play out in court are slimmer.

They're free because they could not achieve network effects if there was a monetary cost involved for users. So basically if they weren't free they wouldn't exist (at scale).

With or without EU regulations, client software could decide to discard all cookies once the user has "left" the site. Or it could block cross domain cookies of its own volition. Yes, it doesn't fix the fundamental issue, but it does address it for those that want to fix it against the tide. Yes, it comes with drawbacks, but it is what it is so long as we don't collectively move towards paying for content, ideally in micro form.

I sometimes come across articles in local publications that ask me to subscribe - dude, seriously? Do you expect me to subscribe to an Alaskan publication when I live half the world away and could not care less of what happens there, but just want to read this one article that seems interesting?

So instead we have ad funded websites that have to do what they have to do in order to make some money and keep publishing whatever it is they publish. Hence tracking cookies.

Everyone's needs would be better served if we could pay for content the same way we did back in the day of printed newspapers. You buy today's edition and you get today's edition and no one except the newsagent is tracking you (if you happen to regularly buy the newspaper from her, she'll remember you, and she may even suggest additional newspapers to buy but it's implied, right? we dislike machine tracking, not humans remembering our buying habits).

Alas, we don't have that. We have intrusive tracking and subscriptions, even though technically it's something we could build in weeks (lest the payment companies didn't make it unfeasible, for their own benefit).

And people do sometimes try to figure it out. Bundles come to mind. Everything -- except micro transactions allowing you to purchase just. this. article. And while micro transactions don't exclude tracking, companies are more likely (is this wishful thinking?) to be careful with a paying customer's experience than with freeloaders, which is what we insist of being, while putting up demands as to what publishers can do with our data.

and so has the internet. some use it for good, others for evil.

these are behaviours and traits of the user, not the tool.

So much of the horribly unmaintainable code I've come across was the result of overengineering that whenever I see an abstraction I have a compulsive urge to scream into the void. Some (the simplest ones) are ok, but other times an abstraction that seems like a good idea on day 0 evolves into an unmanageable hodgepodge of hacks that makes an innocent developer that bumps into it years later question their choice of career.

It's a hybrid 20-ish person team and I work from the office 4 days a week (my choice), but yeah, you're right that we fail at async coordination and fall back onto calls. I've suggested alternatives (meet in the office a couple of days a week) but it has failed to materialise and the only workable solution I found has been to remove myself from noisy/unproductive calls.

I hear what you're saying about docs but knowing everyone I don't feel we could do it. Pre-covid everyone-in-the-office worked well, but you need to drag people into the office these days so it seems we're stuck with an ineffective hodgepodge of calls and in person fractional team chats.

I don't mind working from home, but I find WFH-accommodating calls exhausting; it feels they take far longer than the equivalent in person discussion, in addition to losing the ad-hoc nature of office conversations. The context switch feels heavier on calls than it does in person. Also, the way daily calls tend to be scheduled (everyone at once, let's hash out everything for everyone so we can then get on with things) means I have to sit in on conversations that are irrelevant to me and for some reason having noise coming out of my headphones means I can't focus on anything else. On the opposite end, I'm now sat in the office while colleagues are having a meeting right next to me and I am able to focus on writing this reply (or maybe coding, if I wasn't procrastinating on HN).

Maybe it's my audio setup, but I've tried multiple types of headphones to no avail. I feel the quality of my work (and life) would improve if I could cope better with calls, but despite my efforts, I can't do it. I've removed myself from calls with an adjacent team that mostly don't concern me, and asked them to ping me only if they need me. That saved 30m to 2 hours of my life and saved me from becoming exhausted before the day has properly begun.

There's also a problem in that _some_ people prefer WFH because they can slack off more (or have multiple jobs); that means they're not always available and not fully committed, with whatever implications for the people who rely on them and the company/product.

Maybe it's down to the pace of the work. Startups, I feel, benefit from in office huddling. Corporations with ample time on their hands could be better suited to WFH.

I'm no branding expert but this sounds like a bad idea, yet given Musk's whimsical decisions, I'm not entirely sure he's serious about it rather than trying to reap the benefits of keeping Twitter in the news.